An Evaluation of the Vigil System - New York State ... · Laura Pucher, B.A. Stephanie Silver, M.P.H. Melina A. Soriano, A.A. ... charts. Information regarding date of birth, age,

NYS DOH Dementia Grants Program 2003 Project 1

New York State Department of Health Dementia Grants Program2003-2005 Grant Funded Project

An Evaluation of the Vigil System

Hebrew Home for the Aged at Riverdale5901 Palisade AvenueRiverdale, NY 10471

Ph. 718-581-1741Administrator: Daniel Reingold, MSW, JD

[email protected]


VIGIL Final Report

Project Title: An Evaluation of the VIGIL System.

Leading Nursing Home: Hebrew Home for the Aged at Riverdale.

Project Director: Douglas Holmes, Ph.D.

Names of participating Nursing Homes:

Subcontractors:

Project Staff:

Jeanne Teresi, Ed.D, Ph.D.Mildred Ramirez, Ph.D.

Data Collection/EditingMaria Badri, A.A.Gabriel D. Boratgis, M.P.H.Lucja Orzechowska, Ed.M.Laura Pucher, B.A.Stephanie Silver, M.P.H.Melina A. Soriano, A.A.Ray J. Soriano, B.A.Gail A. Sukha, B.A.Nancy Trujillo

Data EntryPatricia BisonoVirginia MartinezMelina A. Soriano, A.A.Ray J. Soriano, B.A.Nancy Trujillo

Data Processing/AnalysisJulie Ellis, Ph.D., R.N.Joseph P. Eimicke, M.S.Jian Kong, M.S.

AdministrativeMartha Heatley, B.A.


Section I: Goals, Objectives, Research Questions, Hypotheses

This project assessed the extent to which modern technology can augment, and/or

substitute for direct staff intervention in non-acute late evening and night-time situations in the

nursing home setting. The impacts associated with the presence of such technology on the

quality of care provided to, and quality of life experienced by, nursing home residents with

dementing illness are the topics central to this project. A service (VIGIL) was implemented for

residents of a Special Care Unit (SCU) maintained by the lead nursing home for persons with

dementing illness; an SCU matched for cognition was used as a comparison unit on which the

service (VIGIL) was not implemented.

There were three clusters of hypotheses. These related to primary resident and staff

outcomes. (The alpha level for each variable within the cluster was prespecified at 0.05 for a

two-tailed test.) The study hypotheses were:

1. The presence of an automated sensing system (VIGIL) will reduce the number of times

that nurses and Certified Nurses’ Aides (CNAs) have to check residents’ nocturnal status.

2. Due to the presence of VIGIL, the absence of excess noise and light associated with

otherwise mandated routine rounds will enhance sleep quality and permit/promote sleep

consolidation, thus reducing the occurrence of deviant behaviors and improving the affect

of residents.

3. Through the use of the intervention aspect of the VIGIL system, its presence will

contribute to the reduction of falls and accidents and injuries, thus improving quality of

life for the residents.


Section II: Background and Rationale

Day and colleagues1 provide a conceptual orientation to the study of the environment,

discussing the need to consider individual characteristics in the design of environments. This is

reminiscent of the concept of person-environment fit introduced 30 years ago by Kahana2, which

provided a context for the study of human factors that examines the relationship between task

demands and personal components. The environment is important because it is closely linked to

quality of life, defined by Lawton as “the multidimensional evaluation by both intra-personal and

social-normative criteria, of the person-environment system of the individual” (p.6)3.

As reviewed in Holmes, Teresi and Ory4, despite decades of research regarding person-

environment fit, it is still unclear how best to individualize care environments, given the

constraints implicit in the need to serve many individuals, with varying needs, in a single

congregate setting.

Day and colleagues1 emphasize that targeting and tailoring to stage of dementing illness

is an important aspect of environmental design. They show that some environmental

modifications work best for people at moderate stages of impairment, while others work better

for people at later stages. This point cannot be over-emphasized, as the prevalence of cognitive

impairment in chronic care settings is high. For example, the National Institute on Ageing

collaborative studies found that almost 90% of nursing home residents were cognitively

impaired, about half severely so (see Teresi, Morris, Mattis and Reisberg5).

Confusion and disorientation can alternatively be exacerbated or helped by environmental

interventions. An example of both the importance of considering the environment in the context

of the people (residents and staff) and the difficulty associated with attempted remediation is a

study by Schnelle and colleagues6 of the impact of an intervention to reduce noise and light at


night, which they have shown to be associated with awakenings. Staff members were so resistant

to the intervention that they never achieved a noise reduction characterized by episodes below

50db. These findings, showing the difficulty in implementing the most basic of interventions,

highlight the need for sharpened awareness on the part of administrators and regulators regarding

the importance of the environment. Sloane, Mitchell, Calkins and Zimmerman7 document the

lower than recommended levels of lighting found in both activity areas and in resident’s rooms

in SCUs, along with uneven lighting and glare. Additionally, decibel levels in the 60s were found

in the dining room during lunch and outside the nurses’ station, a level equivalent to loud talking.

Given the association of low lighting, circadian rhythms, sleep disorder and agitation observed in

several studies, and the relationship between noise and negative outcomes observed in others,

this situation is less than ideal and deserves attention. It can be posited that the reduction of the

need for night rounding could reduce noise and result in sleep enhancement and improved

behavior and affect.

Many facilities are understaffed: nursing shortages have become endemic and public

reimbursement for institutional care has been reduced. Further, great emphasis has been placed

on reducing or eliminating the use of physical and/or chemical restraints as a form of behavior

management. There has also been a focus on eliminating the use of bedrails for residents, as it

has been shown that they can increase the number of falls and other injuries as residents attempt

to get out of their beds during the night8. Indeed, a decrease in bed rail use did not result in

increased serious injuries for nursing home residents9.

One solution to the need to attend better to resident safety, while maintaining

independence has been the introduction of resident monitoring systems. At the annual meeting of

the American Association of Homes and Services for the Ageing (AAHSA) in 2004, there was a


session focused on “Solutions to the Aging Services Crisis,” in which presentations from

universities and private corporations demonstrated new technologies that have been developed to

enable caregivers to provide services more efficiently, particularly in nursing homes. Experts at

the conference asserted that Skilled Nursing Facilities (SNFs) and assisted living will be the

primary focus of technologic innovation during the next several years, so as to reduce costs and

improve staff efficiency. Fall-monitoring equipment is not currently considered a routine aspect

of nursing care, and knowledge about the value of such devices is poor;10 more evidence is

needed on both the benefits and shortfalls of using monitoring equipment.

A major theme of the 2005 International Association of Homes and Services for the

Ageing Conference (IAHSA) was technology and its uses in aged care, including both the

monitoring of older people in residential care or in their homes, and the use of technologies to

enhance the role of the aged care workers. Glascock and Kutzik11 presented the Automated

Behavioural Monitoring System (ABMS), which has been in use for 10 years. This system uses

off the shelf components. Based on use by 30 home care clients, the developers claim that

alterations in care was achieved for 26 clients, and that the relationship between the nurses and

the clients was enhanced because the clients felt an increase in their ‘peace of mind’. Such

anecdotal evidence underscores the need for more vigorous evaluation of monitoring systems.


Section III: Methods

Description of the Intervention

The VIGIL monitoring system, evaluated in this study, is comprised of a bed exit sensor,

which is positioned under each resident’s bed sheet, and bathroom and bedroom exit monitors.

An incontinence sensor is also available. VIGIL alerts caregivers via a silent pager when a high

risk resident exits their bed, bedroom, or bathroom according to rules that are established for

each resident. This allows caregivers to aid the resident and potentially reduce falls. In addition

VIGIL records caregiver response times. For this study, the incontinence sensor was not used

because nursing staff felt that the diapers used were sufficiently absorbent that the monitor would

not be able to detect moisture.

Study Design

The study was a quasi-experimental design that compared the residents from a unit in

which the VIGIL system was installed (the intervention unit) with residents from a unit in which

there was no such installation (the comparison unit). The VIGIL system was installed in one of

the facilities larger special care units (SCU)(50 beds), which serves residents with moderate to

severe levels of dementia; another 50 bed SCU – in a separate building – matched for case mix

index and cognitive impairment was the comparison unit.

Procedures

In addition to examining the case mix index for the units, the MMSE data used routinely

for placement decisions were considered in the selection of the two units that appeared to be the

best matches among the 17 candidate units.


All residents who lived on the two units were approached to participate in the study. For

those who were not able to give consent, their families were approached to give consent on

behalf of the resident; one family refused to allow contact with the resident.

A computer was installed in the nursing unit and rooms were wired to receive VIGIL.

Additionally, bed pads were ordered. The system was installed free of charge. Nursing staff were

trained in its use; additional retraining was later required. VIGIL staff performed this training

on-site, in addition to ongoing remote monitoring. Research staff showed CNAs how to use the

pagers, and nurses how to set rules for egress. For example, rules could be set so that the CNA

would be paged only after prespecified time in the bathroom. The time would be determined by

the nursing staff, based on previous patterns of behavior. The research assistant helped the

nursing staff set rules, and monitored carefully the behavior of the staff and kept daily logs

regarding implementation (e.g., rules set and bed pad operations); performance feedback was

provided to staff.

Sample

At baseline, on both units, residents were primarily female, white, widowed; 44% of

them had worked in clerical or sales positions. The average age on both units was 87 (±7.5

years). There were no significant demographic differences between the two groups. A total of

118 residents were eligible for baseline, in-person data collection. As new residents replaced

those who had left the unit, they were entered into the study. A total of 92 residents completed

the baseline assessments. Ten residents partially completed the instruments and 16 refused or

were unable to participate due to hospitalization or illnesses.


Data Collection

Data were collected at 3 points: baseline, 12-month follow-up and 15-month follow-up.

A total of 78 residents (38 on the intervention unit and 40 on the comparison unit) completed a

baseline and one additional follow-up, and constituted the analytic sample. Sixty-six residents

completed all three waves of data collection.

Instruments

Prior to in-person data collection, demographic data were obtained during reviews of residents’

charts. Information regarding date of birth, age, place of birth, gender, marital status, education

level, past occupation, and father’s first name were gathered from charts, to be used when

assessing residents’ memory and orientation.

Resident Interview: In-person assessments were conducted with residents during approximately

an hour long session, in which interviewers administered the following assessments: the

Institutional Comprehensive Assessment Referral Evaluation (INCARE)12-20 and the INCARE

Cognitive Screening measures (including Mattis Attention Subscales21, the Range of Motion

Scale, the Feeling Tone Questionnaire22, the Cornell Scale for Depression in Dementia23, and the

Performance Activities of Daily Living (PADL)24. Interviews were conducted in a private

setting, in either the resident’s room or a quiet sitting area located on each unit.

Observed Behavior & Affect: Interviewers performed five minute behavioral observations at

ten different times: morning, afternoon and evening, over a three day period, in order to capture

residents in normal, everyday settings. Observations of resident affect were examined through

the Behavioral Observation Checklist25 and the Observed Emotional Scale26.

Functional Capacity: CNAs were interviewed by research staff member to complete the

Functional Assessment Staging (FAST)27 and the Personal Activities of Daily Living


Questionnaire25. These interviews were conducted on three separate days and followed-up every

three months. They occurred in private, at the nurses’ station located on each unit, and with

consideration to the activities of daily activity schedules for each CNA.

Staff Informant Questionnaire, Behavior & Activities: Social workers familiar with each

resident were asked to report on residents’ affect and sleep patterns and activity participation

using the Multidimensional Observation Scale for Elderly subjects (MOSES)28. Also, therapeutic

activity workers most familiar with each resident were asked to report on residents’ social

contracts and activities.

Sleep Logs: Staff Burden & Direct Care: Facility nursing staff were periodically asked to fill out

weekly log sheets describing sleep behaviors for each resident. Night shift CNAs logged the

amount of time spent caring for residents, and answered questions about resident nighttime

behaviors and sleep interruptions and staff burden. Log sheets were completed one night per

week (approximately) for each resident. At each wave of date, on average, 12 to 30 weeks of

data were collected for each participant.

Falls and Injuries: A Quality Assurance Audit summary was completed by the nursing

department throughout the three waves of data collection, which detailed accidents and incidents,

including falls and internal and external risk factors for falls.

Chart & Medical Record Data on Behavior: The Patient Review Instrument (PRI) contains

four items that reflect resident behaviors: physically abusive, verbally abusive, wandering and

socially inappropriate behavior. These data were used primarily to examine any long-term

changes, which may have occurred accompanying extended exposure to the intervention.

Analytic Variables

Covariates


Covariates were selected based on variables which were significantly different at baseline

between intervention and comparison groups. Although the units were matched initially on case

mix index and cognition, it was not possible to control completely for differences in resident

characteristics.

1. Cognition: The Mini-Mental State Examination (MMSE) was used in these analyses. The

twenty item scale (alpha of 0.75 for this sample) measures the degree of cognitive

impairment, and is scored in the disordered direction.

2. Activity: The social contacts/activities scale from the Helmes and Csapo’s

Multidemensional Observation Scale for Elderly subjects (MOSES)28 (alpha of 0.74 for

this sample) consists of six items that measure how frequently in the past week residents

were “initiating interactions with day shift staff members or residents”, “helping other

residents” or “responding to social contacts”. This scale is scored in the positive

direction.

3. Activities of Daily Living: The Performance Activities of Daily Living (PADL) (alpha of

0.93 for this sample) is a 27-item scale that measures an individual’s lack of ability to

perform certain activities of daily living independently. This scale is scored in the

deviant direction.

4. Age: Age was considered as a covariate because of its theoretical relationship to

outcomes such as falls. As shown below, there were no differences between the units on

average.

5. Walking Outside: The walking outside scale consists of 5 items (alpha of 0.81 for this

sample) that measures an individual’s ability to walk outside of the building. This scale is

scored in the deviant direction.


Outcomes

Prespecified outcomes were:

1) Resident falls and injuries received as an outcome of those falls.

2) Affect: The Feeling Tone Questionnaire (FTQ)22: The FTQ Affect scale contains 16

questions asked of the resident; typical items are: “Are you feeling well?”; “Are you

feeling happy today?”; “Do you feel lonely?”; “Do you have a good appetite?”; “Do you

sleep well?”. Answers to these questions are scored, and additional ratings on affect,

based on the responses are recorded. Resident affect was rated on a 5-point scale which

describes affect as being: “extremely positive,” “moderately positive,” “neutral,”

“moderately negative,” “extremely negative.” The FTQ is collected three times per

administration and an average score is computed from these. The alpha coefficient for the

affect scale for this sample was 0.80. This scale is scored in the deviant direction.

3) Observed Behavior: The Behavior Observation Checklist25 measure is a 37-item

behavioral observation checklist completed by trained raters. Typical items include:

“Argumentative”; “Asking for help”; “Noisy”; and “Uncooperative.” The individual was

rated with respect to each behavior as to whether the behavior occurred “Not at all, ”

“Very little (1x or 2x during observation period), ” “With some frequency (several

times), ” “With moderate frequency (many times but not continuous), ” or “With great

frequency (continuous). ” Residents were observed ten times per administration and an

average score from these observations is computed. The alpha coefficient for this sample

was 0.68 for observed behavior. This scale is scored in the deviant direction.

4) Staff Direct Care, Intervention and Burden: Direct Care was measured using one item,

“Time spent with resident,” (recorded in minutes) from the nurse reported Sleep Logs;


Staff Burden was measured using a scale developed from seven items in the nurse

reported Sleep Logs. Items were scored on either a three or five point Likert scale.

Typical items are: “How well did resident sleep tonight?” (“Very well,” “Well,” “OK,”

“Not very well,” and “Not well at all”); “Did you have to check on resident?”; “Was

resident night-time behavior a bother to you?” (“Not at all,” “Somewhat,” and “A great

deal”); “How cooperative would you say the resident was tonight?”; “Night-time care for

resident is getting?” (“Much harder,” “A little harder,” “About the same,” and “Much

easier”); and “Do you think resident night-time behavior is?” The other items had similar

response categories. Individual items were averaged over a month period, and then the

seven items were summed to create a score for the month. The Cronbach’s alpha for this

measure was 0.92. This scale was scored in the burdened direction.

Rationale for Selection of Outcomes: The most effective fall prevention strategies are

multifactorial, with multiple associated outcomes. It was thus hypothesized that the outcomes

would be multifactorial.

Cluster 1: It was hypothesized that falls, accidents, and injuries would be reduced in the

intervention group.

Cluster 2: Affect was selected as an outcome because residents who have nighttime

interruptions are more likely to feel worse the following day, exacerbated by having to wait for

attention or care. Additionally, falls are related to decreased affect. Thus it was posited that the

intervention group would experience, on average, decreased negative affect. Observed Behavior

was selected as an outcome because it has been shown that disturbed nocturnal sleep was related

to agitation29. It was hypothesized that a decrease in behavior disorder would be observed.


Cluster 3: Direct Care was selected as an outcome as it was expected that the time the nurses

spent with the residents during the night would decrease, if rounding decreased. Staff Burden

was selected as an outcome as it was expected that the use of the VIGIL system would decrease

staff burden. Additionally, it was posited that the residents who were exhibiting disturbing

behaviors during the night would be attended to quickly and cause less interruptions and

therefore less burden for the nurses.

Analytic Procedures

A repeated measures mixed model analysis was performed. The mixed (random and fixed

effects) model is used to adjust for design effects associated with residents clustered within units.

The correlation within the unit can be modeled through consideration of an appropriate

covariance structure (in this case, unstructured). The model holds even with unbalanced and

missing data, so long as the missing data are random. The SAS PROC MIXED software

constructs an objective function associated with Maximum Likelihood (ML) or

Restricted/residual maximum likelihood (REML), and maximizes it over all unknown

parameters.

Generalized linear mixed models have been specially developed for outcomes that are not

normally distributed (in this case, the falls and injuries). It resembles generalized least squares,

the fixed effects component of the mixed model procedure. SAS GLIMMIX was used to model

falls or injuries each month. Further, the robustness of the model was checked with the SAS GEE

procedure, a quasi-likelihood formulation useful in accounting for missing observations and

handling continuous covariates that may be time dependent.

The resident constituted the unit of analysis for the resident outcomes, and there was

clustering within the units. (Our previous analyses of these types of data, and the fact that only


one facility is sampled indicate that the facility design effect can be ignored, but that the unit

effect is important.) Therefore, sample sizes had to be larger to account for unreliability of

measures and for the design features of clustering. Despite the attempt to match units, some

inter-unit differences were observed. For example, the intervention group was more cognitively

impaired than was the comparison group.

Extensive psychometric analyses have accompanied development and cross validation of

the CARE scales as well as most other data collected. CARE scales have retained high reliability

(coefficients in the 80’s and 90’s) for most samples. Validities, depending on the method for

assessment, remain adequate to good. Psychometric analyses were conducted to check the

properties of the scales for this sample.

Preparatory to final bivariate or multivariate analyses, extensive initial analyses were

performed. In the case of nominal/ordinal data, various forms of dummy coding were used.

When appropriate, bivariate data plots and residual plots e.g. box plots, scatter plots, histograms

and normal probability plots were examined in order to detect possible departures from linearity,

heteroscedasticity and outliers. Diagnostic statistics (Cook’s D values, studentized residuals)

were used to assess the influence of individual data points, detection of outliers and examination

of model fit preparatory to analyses; formal colinearity diagnostics were performed.

Adequacy of group matching was assessed with a low threshold (p-values < 0.15) for

detecting deviations from equality of means, variances and proportions in demographics and

baseline variables within the two groups.

The covariates described above were included. Binomial tests were conducted on

dichotomous and Poisson tests were carried out on non-binomial (e.g., count) data. Analysis of

falls and injuries data was performed using up to 15 months of data. If a resident had any fall or


injury in a month, regardless of number, they were coded as having a fall or injury for that

month. The number of falls or injuries per month were not taken into account because there were

few multiple falls or injuries per month. The analyses were first run including age as a covariate;

however, because age was not different between the groups, and was also not significantly

related to the outcomes in the multivariate context, it was removed in the final analysis.


Section IV: Results

Bivariate Analyses of Group Differences in Potential Covariates

Examination of the cognitive scales contained within the INCARE12-25, 30-34 revealed that

at baseline residents exhibited moderate to severe levels of cognitive impairment, including

several non-testable residents. Non-response was due to residents’ lack of alertness or

arousablility, displayed inability to respond to simple commands, and responses offered that

were primarily rambling and/or tangential. Residents of the intervention unit evidenced more

cognitive deficit than their counterparts on the comparison unit. (See Table 1.) Direct resident

assessments indicated that the intervention group displayed more physical limitation in their

range of motion and performance of ADL’s. Additionally, therapeutic activities workers

reported that this group was more likely to attend or participate in social activities and functions.

Table 1: Baseline descriptive statistics for the covariates.UNIT

SP3 (Comparison) MG1 (Intervention)Covariate Mean Std. Dev. Mean Std. Dev. t-test p-value

Age 87.55 7.49 87.43 7.00 0.930

Cognition (MMSE) 16.23 5.01 19.84 7.20 0.005

Activities of DailyLiving (PADL) 3.39 6.62 8.20 9.65 0.010

Activity (ActivityParticipation Scale) 16.45 4.26 19.10 5.58 0.004

Walking Outside 2.84 1.17 3.62 0.85 0.001

Bivariate Analysis of Group Differences in Outcome Variables

Accident/Incident Information

As indicated by the Accident/Incident Reports, a total of 192 accidents occurred through

the duration of our study. The majority of these accidents did not occur during resident transfer


(88.5%), but 46.4 % of all accidents reported occurred in residents’ rooms. Data indicated that

75.5% of accidents were the result of a fall, with 49.0% of those falls the result of an unknown

origin and involved residents being found on the floor. The least number of accidents (1.0%)

occurred at the nurses’ station. Among those who experienced an accident, 53.7% sustained an

injury. Injuries other than lacerations, hematomas, abrasions, etc. were the most common

(37.9%). Lacerations or cuts were the second most noted injury (30.1%). There were no

occurrences of concussions, sprain, strain, or viscera injury reported, and only one reported

occurrence of a burn or scald caused by spilled food or drink (0.5%). The two groups did not

differ significantly in terms of number of falls at baseline; however, there was a significant group

difference in falls at baseline with respect to the final analytic sample, in the context of the

multivariate analyses.

Behavior and Affect

As indicated by nurse informants, the intervention group more prominently displayed the

occurrence of sleep disturbance and disturbing behaviors. Overall, this group was more prone to

exhibiting agitated behaviors, specifically aggressive and physically agitated behaviors. The two

groups did not differ significantly in the occurrence of verbally agitated behaviors. Psychotic

behaviors were also more evident in the intervention group, with an especially higher occurrence

of hallucinations. The intervention group also demonstrated greater baseline affective disorder,

according to observed measures and direct assessment (FTQ) of affect. However, this difference

was not significant in the multivariate analyses. A slightly higher occurrence of anxiety/fear and

sadness was observed from the AARS scores for this group.

Staff Direct Care & Burden


Staff for the intervention group reported a greater amount of direct care at baseline than

did the comparison group staff; however, this difference was not significant in the multivariate

analyses. A significantly greater degree of staff burden was reported by intervention group staff

than by staff in the comparison group.

Examination of the Bivariate Associations Among Variables

As was expected, older residents were significantly more cognitively impaired,

participated in few activities, and had more impairment in walking outside.

Longitudinal Analyses and Tests of Hypotheses

Mixed models analyses were performed in order to examine the outcomes, controlling for

the four designated covariates. Terms for group effect, time, and the group by time interaction

were entered into the model. A significant group term indicates that the groups were different at

baseline. A significant time affect indicates a worsening over time of the outcome, and a

significant interaction term indicates that the intervention group (relative to the comparison

group) improved (declined less) than the comparison group.

The intervention group experienced fewer falls initially (estimate = -1.61, p = 0.017), and

received a significantly higher Staff Burden score from the nurses at baseline, than did the

comparison group (estimate = 3.29, p < 0.001).

Examination of the time effects, from baseline to 15 months indicated that there was no

significant change in the number of falls over the project period (estimate = -0.07, p = 0.148),

and no significant change in the rates of injuries (estimate = -0.03, p = 0.645). The amount of

direct care decreased significantly overall (across comparison and intervention groups) over the

project period (estimate = -0.31, p < 0.001).


A test of the hypotheses that the intervention group, as contrasted with the comparison

group, would evidence a reduction in disturbing behavior, affect, and falls and injuries over time

was accomplished through examination of the interaction of group*time.

The odds of falling over the 15 month period was not significantly different (estimate =

0.09, p = 0.180) between the intervention group and the comparison groups.


Table 2: Results for primary outcomes of the VIGIL system study.

FALLSa

(n = 76)INJURIESa

(n = 76)AFFECTb

(n = 73)BEHAVIORb

(n = 73)DIRECT CAREc

(n = 124)STAFF BURDENc

(n = 124)

Effect EstimateStd.

Error Sig. EstimateStd.





Error Sig.

Intercept -3.5422 1.1877 0.0053 -4.2151 1.5947 0.0125 36.8558 5.1729 <0.0001 5.8599 1.3850 <0.0001 2.9700 4.8631 0.5440 13.3349 1.2483 <0.0001

MMSEPRO 0.0088 0.0287 0.7598 -0.0015 0.0427 0.9721 -0.1669 0.1507 0.2721 0.0286 0.0398 0.4750 0.3072 0.1322 0.0205 0.0233 0.0338 0.4920

PTOTALPR 0.0373 0.0208 0.0728 0.0589 0.0291 0.0436 0.0471 0.1302 0.7185 0.0061 0.0332 0.8538 0.2543 0.1097 0.0208 0.0758 0.0281 0.0073

ACTPARPR 0.0672 0.0423 0.1130 0.0608 0.0581 0.2961 -0.1714 0.1820 0.3498 -0.0361 0.0494 0.4681 0.0867 0.1850 0.6395 -0.0736 0.0472 0.1196

WALKOUPR 0.1419 0.1511 0.3480 -0.0035 0.2071 0.9867 0.8551 0.7655 0.2680 -0.3236 0.2026 0.1149 1.5750 0.6942 0.0236 0.2042 0.1764 0.2474

GROUP -1.6056 0.6686 0.0167 -0.2943 0.7726 0.7034 2.3565 2.0563 0.2559 -0.9558 0.5698 0.0981 -2.4640 1.8641 0.1868 3.2850 0.5176 <0.0001

TIME -0.0674 0.0465 0.1475 -0.0259 0.0561 0.6445 0.0346 0.0932 0.7118 -0.0486 0.0289 0.0969 -0.3127 0.0628 <0.0001 -0.0044 0.0297 0.8824

GROUP*TIME 0.0931 0.0693 0.1799 0.0162 0.0745 0.8280 -0.2864 0.1323 0.0341 0.0363 0.0402 0.3704 0.5292 0.0808 <0.0001 -0.0020 0.0383 0.9583

VARIABLES: D1: AGE OF RESPONDENTMMSEPRO: MMSE PRORATED SCORE (DEVIANT)PTOTALPR: PADL TOTAL: PRORATED (DEVIANT)ACTPARPR: ACTIVITY PARTICIPATION: PRORATED (POSITIVE)WALKOUPR: WALKING OUTSIDE: PRORATED (DEVIANT)GROUP: GROUP (0 = SP3, CONTROL; 1 = MG1, INTERVENTION)TIME: TIME IN MONTHS (FOR AFFECT AND BEHAVIOR, TIME CODED AS 0 = BASELINE; 12 = 12

MONTH FOLLOW-UP; 15 = 15 MONTH FOLLOW-UP)

a Binary outcome , generalized linear mixed model, random on id.b Repeated measures mixed model with unstructured covariance structure.c Means within a month, Mixed model, random on id.


There was no significant differences between the group in rates of injuries (estimate =

0.02, p = 0.828).

The intervention group improved significantly in affective disorder by the end of the

project (estimate = -0.29, p = 0.034).

There was no difference in rates of behavior change between the two groups over time

(estimate = 0.04, p = 0.370).

The staff logs showed that the intervention group received significantly more direct care

over the duration of the project than did the comparison group (estimate = 0.53, p < 0.001). The

additional care was about six minutes per month per resident. There was no significant difference

in staff-reported burden between the groups over time (estimate = -0.002, p = 0.958).

An additional analysis was performed to replicate earlier findings35 showing a

relationship between direct care received on special care units for individuals with dementia and

positive resident outcomes. Examination of an interaction term for direct care received over time

and group status showed that additional time spent in direct care in the intervention group was

associated with decreased affective disorder, thus replicating the earlier findings. (See Table 3.)


Table 3: Predicting Affective Disorder including Direct Care.

Affecta

(n = 67)

Effect Estimate Std. Error Sig.

Intercept 37.3048 5.0223 <0.0001

MMSEPRO -0.0409 0.1480 0.7833

PTOTALPR 0.1411 0.1335 0.2949

ACTPARPR -0.2503 0.1614 0.1263

WALKOUPR -0.1004 0.7436 0.8930

GROUP 11.0304 3.5987 0.0033

TIME 0.2343 0.0580 0.0002

GROUP*TIME -0.4224 0.1305 0.0020

DIRECT CARE*GROUP -0.4498 0.2109 0.0372

VARIABLES: DIRECT CARE: TIME SPENT CARING FOR RESIDENT (MINUTES)D1: AGE OF RESPONDENTMMSEPRO: MMSE PRORATED SCORE (DEVIANT)PTOTALPR: PADL TOTAL: PRORATED (DEVIANT)ACTPARPR: ACTIVITY PARTICIPATION: PRORATED (POSITIVE)WALKOUPR: WALKING OUTSIDE: PRORATED (DEVIANT)GROUP: GROUP (0 = SP3, CONTROL; 1 = MG1, INTERVENTION)TIME: TIME IN MONTHS (FOR AFFECT AND BEHAVIOR, TIME CODED AS 0 = BASELINE;

12 = 12 MONTH FOLLOW-UP; 15 = 15 MONTH FOLLOW-UP)a Repeated measures mixed model with unstructured covariance structure.

DISCUSSION

The main findings were that the groups did not differ in terms of the odds of falling or

rates of injuries over the study period. While there was no difference in observed behavior,

resident-reported affective disorder decreased significantly in the intervention unit. Results of the

analysis showed that the nursing staff spent significantly more time in direct care for the

intervention group than did the nursing staff caring for the comparison group.

The reported staff-burden was significantly higher at the beginning of the project for the

nurses caring for the intervention group; however, this burden on the nurses did not get worse


over the time of the project, and there was no significant difference in burden between the

groups. Although speculative, it is possible that use of the equipment and/or heightened

awareness of and attention to resident needs via the paging system translated into somewhat

more direct care, but no appreciable additional burden. The additional care and possible

vigilance may have been associated with the increased quality of life as measured by affective

status. An examination of the correlations of staff direct care and affective disorder lend support

for this speculation. Correlations of affective disorder and direct care were inversely correlated

across monthly measurement occasions. The strength of the association was greater (and

significant) at the third wave of data; the multivariate findings were that the more time spent in

direct care on the intervention unit, the less affective disorder was reported by residents.


Section V: Strengths and Limitations

A limitation of the study is that nursing staff were resistant to implementing the

intervention. Because the intervention was implemented on a pre-existing unit, with established

routines, staff was reluctant to alter their methods of care. A goal of the project was to reduce

the necessity for frequent night rounding, often up to once every half an hour. The rationale for

the proposed reduction was that such rounding results in the intrusion of light and noise, and

therefore, interrupted sleep, as shown by Schnelle and colleagues6. Like Schnelle and

colleagues, we were unable to change staff behaviors to the extent desirable. Consistent with the

prior protocol for night care, staff continued to check on residents every two hours, and recorded

this on the Sleep Logs. Additionally, research staff was required to monitor the implementation

and to assist in setting rules for egress because nursing staff did not wish to take ownership of

this task.

An additional limitation is a result of the quasi-experimental design; some imbalance

occurred between the intervention and comparison groups in terms of baseline characteristics. It

was not possible to randomize within units because of contamination, and the cost of wiring two

units so that enough subjects could be enrolled.

A strength of the study was that staff received training in VIGIL, and heightened

awareness of risk factors for falls. This may have lead to increased time spent with residents, and

thus to enhanced quality of care and quality of life, as evidenced by a reduction in affective

disorder (improved affect). This conforms with other findings that enhanced staffing and staff

time spent in direct care is significantly related to positive outcomes35.


Section VI: Conclusions

In conclusion, the findings related to VIGIL are generally mixed. There was no

significant, sustained reduction in falls and injuries, but there was a significant difference in

affective disorder in the intervention group as contrasted with the comparison group. There was

no significant increase in staff-perceived burden, despite the significant increase in the amount of

direct care time logged for the intervention group, as contrasted with the comparison group.

Possibly, as a result of training related to monitoring residents, the direct care time logged by

nursing staff increased. It is possible that the associated increase in affect was related to this

increased attention. However, the question remains as to whether the effect was due to VIGIL or

vigilance. If the latter, then increased staff time, provided in the context of training and

monitoring, may be the important ingredient in the enhanced quality-of-life. Because each

hypothesis was treated as a cluster with a pre-specified p value of 0.05 (two-tailed), the

possibility of chance findings was reduced greatly, however, as with all research this is always a

consideration. An aim for future research involves further examination of the relationship

between the intervention, time spent in direct care, and affective disorder.


Section V: Final Budget Report


References:

1. Day, K., Carreon, D., & Stump, C. (2000). The therapeutic design of environments forpeople with dementia: a review of the empirical research. The Gerontologist, 40, 397-416.

2. Kahana, E. (1975). A congruence of person-environment interaction. In T.O. Byerts,M.P. Lawton & J. Newcomber (Eds.), Theory Development in Environments and Aging.Washington D.C.: Gerontological Society.

3. Lawton, M.P. (1991). A multidimensional view of quality of life. In J.E. Birren, J.E.Lubben, J.C. Rowe, D.E. Deutchman (Eds.), The concept and measurement of quality oflife in the frail elderly (pp. 3-27). New York: Academic Press.

4. Holmes, D., Teresi, J., & Ory, M. (2000). Editors, special edition dealing solely withdementia special care units. Research and Practice in Alzheimer's Disease, Vol 4.

5. Teresi, J., Morris, J.N., Mattis, S., & Reisberg, B. (2000). Cognitive impairment amongSCU and non-SCU residents in the United States: Prevalence estimates from the NationalInstitute on Aging Collaborative studies of Special Care Units for Alzheimer's Disease.Research and Practice in Alzheimer's Disease, 4, 117-138.

6. Schnelle, J.F., Alessi, C.A., Al-Samarrai, N.R., Fricker, R.D., & Ouslander, J.G. (1999).The nursing home at night: Effects of an intervention on noise, light and sleep. Journal ofthe American Geriatrics Society, 47, 430-438.

7. Sloane, P.D., Mitchell, C.M., Calkins, M., & Zimmerman, S.I. (2000). Light and noiselevels in Alzheimer's Disease Special care Units. Research and Practice in Alzheimer'sDisease, 4, 241-249.

8. Todd, J., Ruhl, C., & Gross, T. A. (1997). Injury and death associated with hospital bedsiderails: Reports to the US Food and Drug Administration from 1985 to 1995. AmericanJournal of Public Health, 87, 1675-1677.

9. Myint, S., Neufeld, R., & Dunbar, J. (1999). Removal of bedrails on a short-term nursinghome rehabilitation unit. The Gerontologist, 39(5), 611-614.

10. Brownsell, S.H. (2004). Fall detectors: do they work or reduce the fear of falling?Housing, Care and Support, 7(1), 18-24.

11. Glascock, A. P. & Kutzik, D. M. (2005). Practice and Policy Implications of AutomatedBehavioural Monitoring System. Paper presented at the 6th International Conference ofthe International Association of Homes and Services for the Ageing. Trondheim Norway27-29th June.


12. Gurland, B., Kuriansky, J., Sharpe, L., Simon, R., Stiller, P., & Birkett, P. (1977-78). TheComprehensive Assessment and Referral Evaluation (CARE): Rationale, developmentand reliability. International Journal of Aging and Human Development, 8(1), 9-42.

13. Gurland, B.J. & Wilder, D.E. (1984). The CARE interview revisited: Development of anefficient, systematic clinical assessment. Journal of Gerontology, 39, 129-137.

14. Copeland, J., Kelleher, J., Kellet, J., Gourlay, A., Gurland, B., Fleiss, J., & Sharpe, L.(1976). A semi-structured clinical interview for the assessment of diagnosis and mentalstate in the elderly: The Geriatric Mental State Schedule, I, Development and reliability.Psychological Medicine, 6, 451-459.

15. Teresi, J., Golden, R., Gurland, B., Wilder, D., & Bennett, R. (1984). Construct validityof indicator-scales developed for the Comprehensive Assessment and Referral Evaluationinterview schedule. Journal of Gerontology, 39(2), 147-157.

16. Gurland, B., Fleiss, J., Goldberg, K., Sharpe, L., Copeland, J., Kelleher, M., & Kellet, J.(1976). A semi-structured clinical interview for the assessment of diagnosis and mentalstat in the elderly: The Geriatric Mental State Schedule, II, A Factor Analysis.Psychiatric Medicine, 6, 451-459.

17. Teresi, J., Golden, R., & Gurland, B. (1984). Concurrent and predictive validity ofindicator-scales developed for the Comprehensive Assessment and Referral Evaluationinterview schedule. Journal of Gerontology, 39(2), 158-165.

18. Golden, R., Teresi, J., & Gurland, B. (1984). Development of indicator-scales for theComprehensive Assessment and Referral Evaluation interview schedule. Journal ofGerontology, 39(2), 138-146.

19. Gurland, B., Golden, R., Teresi, J., & Challop, J. (1984). The Short Care: An efficientinstrument for the assessment of depression, dementia and disability. Journal ofGerontology, 39(2), 166-169.

20. Golden, R. Teresi, J., & Gurland, B. (1983). Detection of dementia and depression caseswith the Comprehensive Assessment and Referral Evaluation interview schedule.International Journal of Aging and Human Development, 16(4), 242-254.

21. Reisberg, B., Ferris, S., Steinberg, G., Shulman, E., de Leon, M., & E. Sinaiko(1989). “Longitudinal Study of Dementia Patients and Aged Controls: An Approach toMeterological Issues: in Special Research Methods for Gerontology, M.P. Lawton andA.R. Herzog (eds), Amityville, NY, Baywood Pub.195-231.

22. Toner, J.A., Gurland, B., & Tirumalasetti, F. (1999). The Feeling ToneQuestionnaire: Reliability and validity of a direct patient assessment screening instrumentfor detection of depressive symptoms in cases of dementia. Journal of ClinicalGeropsychiatry, 5, 63-78.


23. Alexopoulos, G.S., Abrams, R.C., Young, R.C., & Shamoian, C.A. (1988). Cornell Scalefor Depression in Dementia. Biological Psychiatry, 23, 271-284.

24. Katz, S., Ford, A.B., Moskowitz, R.W., Jackson, B.A., & Jaffee, M.W. (1963). Studies ofillness in the aged. The Index of ADL: A standardized measure of biological, psychosocial function. Journal of the American Medical Association, 185, 94ff.

25. Teresi, J.A., Holmes, D., Benenson, E., Monaco, C., Barrett, V., Ramirez, M., & Koren,M. (1993). A primary care nursing model in long-term care facilities: Evaluation ofimpact on affect, behavior, and socialization. The Gerontologist, 33, 667-674.

26. Lawton, M.P., Van Haitsma, K., Perkinson, M., & Ruckdeschel, K. (1999). Observedaffect and quality of life in dementia: Further affirmations and problems. Journal ofMental Health and Aging, 5(1), 69-81.

27. Reisberg, B. (1988). Functional assessment staging (FAST). PsychopharmacologyBulletin, 24(4), 653-655.

28. Helmes, E., Csapo, K.G., & Short, J.A. (1987). Standardization and validation of theMultidimensional Observation Scale for Elderly Subjects (MOSES). Journal ofGerontology, 42(4), 395-405.

29. Kutner, N.G. & Bliwise, D.L. (2000). Observed agitation and the phenomenon of sundowning among SCU residents. Research and Practice in Alzheimer s Disease.

30. Mann, A.H., Wood, K., Cross, P., Gurland, B., Schreiber, P., & Haefner, H. (1984).Institutional care of the elderly: a comparison of the cities of New York and London andMannheim. Social Psychiatry, 19, 1-6.

31. Blessed, G., Tomlinson, B., & Roth, M. (1968). The association between quantitativemeasures of dementia and senile changes in the cerebral grey matter of elderly subjects.British Journal of Psychiatry, 114, 797-811.

32. Kahn, R., Goldfarb, A., Pollack, M., & et al. (1960). The relationship of mental andphysical status in institutionalized aged persons. American Journal of Psychiatry, 117,326-328.

33. Folstein, M.F., Folstein, S.E., & McHugh, P.R. (1975). Mini-Mental State: A practicalguide for grading the cognitive state of patients for the clinician. Journal of PsychiatricResearch, 12, 189-198.

34. Pfeiffer, E. (1975). A short portable mental status questionnaire for the assessment oforganic brain deficit in elderly patients. Journal of the American Geriatrics Society, 22,433-444.


35. Homes, D., Teresi, J., & Kong, J. (2000). Service inputs and costs of care related tooutcomes among cognitively impaired nursing home residents. The Journal of MentalHealth Policy and Economics, 3, 121-127.

Documents

An Evaluation of the Vigil System - New York State ... · Laura Pucher, B.A. Stephanie Silver, M.P.H. Melina A. Soriano, A.A. ... charts. Information regarding date of birth, age,